[Ffmpeg-devel-irc] ffmpeg.log.20150302

burek burek021 at gmail.com
Tue Mar 3 02:05:01 CET 2015


[00:15] <active8> hi. i had two dvds from two different record sessions using an old sony setup that spit out dir/files SONY_DVD_RECORDER_VOLUME/VIDEO_TS/VTS_01_[12345].VOB where the numbers inside the [] are the 5 files that contain info (shell filename syntax.) Last week, the first two wouldn't play in vlc, IIRC. This week, only the first doesn't play. they each have 3 streams, video audio and data, but they're never in the same order - stream 0 might
[00:15] <active8> be video for one file, but audio for another and no logical pattern to that between the two DVDs. I was trying to concat them and convert to mp4. FFmpeg didn't like it at all. Handbrake did it with no prob. heres the output of ffmpeg -i file for week one and two. http://pastebin.com/raw.php?i=xwaAmAdT
[00:18] <active8> I'd like to know how handbrake did it without any special settings that I made and what I can do to get ffmpeg to handle it - i used cat /media/SONY_DVD_RECORDER_VOLUME/VIDEO_TS/VTS_01_[12345].VOB | nice $ffbin -i - -s 352x480 -c:v libx264 -c:a libfdk_aac outputname.mp4 -- $ffbin is just a shortcut to my ffmpeg latest build
[00:22] <BtbN> Why do you pipe the file into ffmpeg, rahter than just letting ffmpeg open it itself?
[00:23] <active8> piping 5 files from cat to concatenate them. same reesults just doing individual files one at a time, except ffmpeg stops after the first failure
[00:24] <BtbN> I don't think you can't concat vob files that way.
[00:24] <BtbN> *can
[00:24] <active8> someone does cause i found it "out there"
[00:24] <active8> doesn't mean they''re right, of course
[00:25] <BtbN> might happen to work if all files are somewhat identical
[00:27] <active8> even so, ffmpeg doesn't like the first two files, for either week (see output messages) and yeah, maybe (if "somewhat identical" means the order of the streams) then maybe handbrake is mapping them. I wondered if it was a mapping problem - something I might need help with]
[00:28] <active8> what are those files considered? chapters?
[00:30] <active8> handbrake doesn't think so
[00:31] <active8> i mean it doesn't think this: /media/SONY_DVD_RECORDER_VOLUME/VIDEO_TS/VTS_01_1.VOB: Invalid data found when processing input
[00:43] <msmithng> does anyone happen to know what increments size is referring to with -thread_message_queue size (input) ?
[00:43] <msmithng> are we talking bytes, or bits or what?
[00:44] <msmithng> further, what does it default to?
[00:52] <c_14> >This option sets the maximum number of queued packets when reading from the file or device.
[00:52] <c_14> so probably in packets
[00:54] <c_14> defaults to 8
[04:28] <travis_> can I record desktop with no monitor attached?
[04:29] <c_14> If you can start an X session, probably
[04:35] <travis_> I have a cctv system and can view the cams over Internet explorer (inside virtual box), I want to set up a computer that has cams open open in IE and set ffmpeg to record the desktop, with monitor unhooked. I found a script that should record the desctop on a rolling basis, I then want to sync this record folder to dropbox... just to double check.. you think the ffmpeg part of this should work? I can post the script if that helps
[04:35] <klaxa> virtual box has recording built in already, no?
[04:35] <c_14> You can't intercept the cam feed with ffmpeg directly?
[04:36] <klaxa> that would be even cleaner
[04:37] <travis_> can it record on a rolling basis? ie, record say a 5 min clip to a folder and continiously overright to the folder, then sync that folder to drobox?
[04:37] <c_14> "continuously overwrite" ?
[04:38] <c_14> look at the segment muxer
[04:38] <travis_> yes, like record a 5 min. clip then overwight that when its finished?
[04:39] <c_14> yep
[04:39] <travis_> k thanks for the help. I will look into that. Thanks!
[04:39] <c_14> segment muxer with a segment_wrap of 1
[04:40] <travis_> k cool Thanks!
[07:55] <MattCodez> Hi, any idea what "ffmpeg exited with code 127" could mean?
[08:14] <Enjolras> hi
[08:15] <Enjolras> can an AVFormatContext be used for both input and output ?
[09:00] <Mavrik> Enjolras, it has to be :)
[09:00] <Mavrik> but not the same context.
[09:01] <Enjolras> right. That was my question. It's what i assumed because there is one set of stream, but i was mislead by the fact the struct contains two pointers, one for an input format and another for an outputformat
[09:01] <Enjolras> thanks
[09:09] <fffan> does ffmpeg has a funciton to check the input filename is valid or invalid
[09:09] <fffan> check before avformat_open_input
[09:57] <travis_> can anyone provide a basic command that will screen capture to a good quality video file for 5min then to another file with a similar name after 5 min in the same directory, then to a third file with a similar name after that 5 min. and then restart the process automatically at the end, thus overwriting the original files in a looping pattern until I stop it?
[10:04] <retard> can you pay for it?
[10:10] <travis_> No I can not..  If I could I would though.. I know how to do a basic high quality screen record with this:   ffmpeg -video_size 1920x1080 -framerate 30 -f x11grab -i :0.0 -c:v libx264 -qp 0 -preset ultrafast -profile:v high444 capture.mkv  and I know -y is the overwrite option, If not I guess I can do enough googling to put 2 and 2 together
[10:20] <travis_> Sorry if this is a bit too much of a request. I just thought somebody might have the experience and will to do it here on the IRC. I'm a young father with not a lot of time on my hands for yet more days of Linux research to get my home CCTV system recorded to a folder that I can then sync to Dropbox as my $500 Swann CCTV  system can not do the basic feature of cloud backup (if someone steals my DVR im hosed.. I can only watch my CCTV
[10:20] <travis_>  stream on my comp. I figured this would be a workaround for me before my wife rings my neck for wasting $500 on a useless system.. I was hoping Linux would save the day this time.
[10:22] <travis_> At this point i'm out of funds or I would surely pay someone for the help..
[10:23] <retard> it's a fine request
[10:25] <retard> i would probably just stick it in a shell script loop, experiment with sleep values until i get a couple of seconds of actual overlap
[10:25] <retard> but it takes time and effort to experiment, and is a pretty narrow usecase
[10:25] <retard> it's a lot like an extremely long hls setup
[10:26] <retard> i mean an hls setup with a very long segment size
[10:27] <travis_> ok, I found a shell script but it returns errors and I am in the dark with what these commands equate too. the script looks like this: #!/bin/bash
[10:27] <travis_> while true; do
[10:27] <travis_> 	ffmpeg -f x11grab -s
[10:27] <travis_> 1920x1080 -r 60 -i :0.0 -f avi /
[10:27] <travis_> tmp/rolling.avi &> /dev/null &
[10:27] <travis_> 	disown
[10:27] <travis_> 	PID=$!
[10:27] <travis_> 	Sleep 30;
[10:27] <travis_> 	kill -kill $PID
[10:27] <travis_> 	cp /tmp/rolling_1.avi /tmp/
[10:27] <travis_> rolling_2.avi
[10:27] <travis_> 	cp /tmp/rolling.avi /tmp/
[10:27] <travis_> rolling_1.avi
[10:27] <travis_> 	rm /tmp/rolling.avi
[10:28] <retard> looks okay, ther should be a "done" at the end
[10:29] <__jack___> travis_: use mv, not cp
[10:29] <__jack___> travis_: use ffmpeg -t switch for duration, instead of &, sleep & kill (something like ffmpeg -i blabla -t 00:05:00 blabla output)
[10:30] <retard> that depends on whether losing frames is a problem or not though
[10:31] <retard> i would probably use both the -t switch for duration as well as backgrounding the process
[10:31] <retard> with & at the end
[10:33] <travis_> yes sorry it does have "done" at the end but I get this error: ---> Missing argument for argument 's' <--- followed by --->  Error Splitting the argument list: Invalid argument <--- on the next line. (This is without your suggestions made to the script) would anybody mind showing me how you think the script should look?
[10:34] <retard> looks like you have newlines in the ffmpeg command?
[10:34] <travis_> I don't mind loosing some frames
[10:36] <retard> start with inserting your own ffmpeg command, either all on one line or with \ at the end so the shell knows the line continues
[10:38] <travis_> ok I got what your saying I think.. each line should have a \ at the end so it knows its a new line?
[10:38] <travis_> or just enter it all as one line..
[10:39] <retard> \ at the end means that the options continue as if it were all on one line
[10:39] <travis_> k
[10:40] <retard> no \ at the actual last line though
[10:40] <travis_> ok that makes sense
[11:20] <retard> godspeed
[12:20] <m3gab0y> hey guys, I'm trying to squeeze out the maximum efficiency for the h264 encoder and be able to get decent quality at 500 kbps but can't get the optimal settings for some reason, care to help me on this?
[12:20] <__jack___> pastebin the cmd ?
[12:27] <m3gab0y> http://pastebin.com/cZ9WQvSy
[12:35] <m3gab0y> anything? I can see some people getting even better quality at 400 kbps... what am I doing wrong here?
[12:36] <__jack___> m3gab0y: well, you can: change the preset, use 2 pass
[12:36] <__jack___> use crf instead of constant bitrate
[12:37] <__jack___> (use x265!)
[12:37] <m3gab0y> here is the problem - with CRF bitrate is too variable and average is way above 500 in all cases - I'm limite to 512 kbps, also x265 is not supported on my device so it's out of the question
[12:38] <__jack___> can't you use crf with maxrate ?
[12:38] <m3gab0y> about the preset - what presets do you recommend - I wasn't able to find anything that works better than this one, and you can give me an example of 2-pass setup
[12:38] <__jack___> anyway, the preset is a good start
[12:39] <m3gab0y> no, maxrate not taking effect
[12:39] <m3gab0y> even if it's set to 400 it can go all the way up to 1500 at times
[12:39] <m3gab0y> and make the quality worse
[12:39] <__jack___> preset: a slower preset mean better quality / size ratio (= smaller filesize, or better quality), see https://trac.ffmpeg.org/wiki/Encode/H.264#a2.Chooseapreset
[12:40] <__jack___> for the 2 pass: https://trac.ffmpeg.org/wiki/Encode/H.264#Two-PassExample : it will first do a fast lookup of the file, to see where it can be heavily compressed, and where is cannot be; the 2nd pass do the real encoding, using the 1st result
[12:42] <m3gab0y> this i gotta see - i wasn't able to get it working with live streaming anyways.... but will  try again
[12:42] <__jack___> hum, the "http" is a live ? hum, forget 2-pass then
[12:43] <m3gab0y> yes, it's live input and i'm also doing live output (transcoding)
[12:44] <__jack___> hum hum
[12:48] <m3gab0y> OK so basically how can we optimize those settings that I have at the moment? Any suggestions?
[12:50] <__jack___> preset, if you have enough power to keep the flow realtime
[12:52] <m3gab0y> it's at the medium preset and I've tested veryslow - almost unnoticeable difference  but 4x the CPU usage
[15:20] <poste9> did we have any recent changes on AVFormatContext::interrupt_callback ?
[15:20] <poste9> looks like it was a pointer, now its not.
[15:21] <poste9> at least Im trying to use it on avio_open2 and its not working because of it
[15:25] <poste9> omg, just ignore what I said.
[15:34] <_genuser_> so to merge two h264 files, they must first be decoded to mpeg?
[15:34] <_genuser_> merge and then encoded?
[15:35] <Mavrik> that's a very old guide
[15:36] <Mavrik> now we have concat filters
[15:36] <Mavrik> but they probably do have to be reencoded yes
[15:36] <_genuser_> https://trac.ffmpeg.org/wiki/Concatenate
[15:36] <Mavrik> https://trac.ffmpeg.org/wiki/Concatenate#filter
[15:36] <_genuser_> that's the one I was looking at it, does talk about using concat filters...
[15:38] <_genuser_> lol. more than the actual concatenation, I marvel at the skill of the coder who parses all those command line params.
[15:39] <_genuser_> Mavrik: so based on the concat filter example. I should be able to pull out chunks of the show (leaving out commercials), and then concat them together.
[15:40] <_genuser_> they'll be all h264 and same frame size, etc. you think that would require re-encoding?
[15:40] <Mavrik> if they would be part of the same stream, then no
[15:40] <Mavrik> you could probaby get away with using the concat demuxer then
[15:41] <_genuser_> is there GUI that can allow you to mark in/out of segments and then uses ffmpeg to pull them out and concat?
[15:41] <_genuser_> or would that have to be written?
[15:42] <_genuser_> and I'd also have to compile the latest ffmpeg probably.
[15:42] <_genuser_> latest I have was built June 13, 2014
[15:43] <Kolizer> hello, how grab window id on ffmpeg?
[15:47] <_genuser_> Mavrik: thanks for
[15:47] <_genuser_> mentioning the demuxer. trying that first.
[15:52] <_genuser_> geez -ss index -t length pulls out a video/audio horribly out of sync.
[15:52] <msmithng> thanks c_14, I didnt see your reply to my question about the thread_message_queue last night.
[15:53] <Kolizer> help, how grab window id on ffmpeg?
[16:16] <spagoop_> hi friends, does anyone have experience using the "blend" effect?
[16:16] <spagoop_> is it possible to achieve something like this: https://www.youtube.com/watch?v=9p0TdX4iizs
[16:16] <spagoop_> ?
[16:17] <spagoop_> for context: this is a part of a bot that will be uploading these videos to youtube every 30 or so minutes
[16:47] <luc4> Hello! I'm using ffmpeg to extract frames from videos. I read ffmpeg seems to support vaapi. Should I be able to reach better performance doing this if I compiled ffmpeg with vaapi on Intel HD?
[16:50] <poste9> is avformat_write_header supposed to leak memory?
[16:53] <BtbN> ffmpeg does not automagicaly use vaapi acceleration. You have to do quite a lot of work yourself.
[16:54] <BtbN> ffmpeg basicaly only does the de-slicing and fills the vaapi input buffers.
[16:56] <luc4> BtbN: I see, so the output buffer from the decoder is not used in any way, correct?
[17:02] <Mavrik> poste9, only if you forget to call avformat_write_trailer and close the format :)
[17:04] <poste9> omg u.u thanks
[17:17] <poste9> well.. now I created a frame with avpicture_alloc((AVPicture *)blackframe, pixfmt, w, h) and never used it, so in the end I do av_frame_free(blackframe); And valgrind was complaining about possible leak, then I put av_free(blackframe->data[0]) and valgrind shuts up... Is that a bug or valgrind is just too strict?
[17:20] <Mavrik> ugh
[17:21] <Mavrik> avpicture_alloc is a hugely outdated API
[17:21] <Mavrik> which includes av_free
[17:21] <Mavrik> use av_frame_alloc, av_frame_get_buffer and av_frame_free
[18:10] <_genuser_> Mavrik: if trimming with ffmpeg is causing a/v sync issues, it's due to not being cut a keyframes etc. and codec related?
[18:10] <Mavrik> it COULD be
[18:10] <Mavrik> but probably isn't
[18:11] <Mavrik> hard to say :/
[18:11] <_genuser_> but it's not being caused by ffmpeg, correct?
[18:12] <_genuser_> I normally separate them out and re-merge them with offset. but I think for an entire video that makes sense. if I'm gonna cut bits out to leave out commercials and then do that for each bit, that's a bit much.
[18:12] <_genuser_> I should be edit before converting to hx64
[18:12] <_genuser_> *h264
[18:14] <poste9> omg..  when I removed avpicture_alloc it removed a lot of valgrind messages, like 15
[18:15] <poste9> thhanks mavrik
[18:33] <poste9> can you guys give me something to look for about the align parameter the av_frame_get_buffer ? I want to understand why everyone use 32.
[20:12] <amalia> Congrats michaelni ffmpeg was selected for GSoC 2015
[20:31] <slaan> Hi !
[20:34] <slaan> Did you know if we can do something like "between(t,0,"length/2")
[20:34] <slaan> as (main_h-overlay_h)/2 with the height ?
[20:35] <jthemovie> Hi Everyone
[20:36] <jthemovie> Could anoyone please tell me why since i update ffmpeg to v2.5.4  i can't download any HLS stream running on SSL ?
[20:36] <jthemovie> A TLS fatal alert has been received.
[20:36] <c_14> slaan: probably, what's your full commandline?
[20:36] <c_14> jthemovie: same compilation flags?
[20:37] <jthemovie> well nope, it was an already compiled version
[20:37] <jthemovie> so i guess the default version enable-ssl check right?
[20:38] <jthemovie> --enable-openssl , is it this one?
[20:38] <slaan> Hi c_14 ! My command work with overlay=10:10:enable='between(t,0,30). but i want to do the same thing with the half of length
[20:38] <c_14> jthemovie: that or --enable-gnutls
[20:39] <slaan> with many video with different size
[20:39] <jthemovie> thanks c_14, so all i need is which options launching ffmpeg disable it
[20:39] <jthemovie> like ffmpeg ignore-tls
[20:39] <jthemovie> im sure there is such option but i couldn't find anywhere :/
[20:40] <c_14> slaan: you can use main_w or overlay_w, yes
[20:40] <slaan> yes, for the space, but for the time ?
[20:41] <jthemovie> could you help me out with it please or do i have to recompile the whole ffmpeg ?
[20:41] <slaan> for example, if my video is 2min, i want to overlay only the first minute
[20:42] <c_14> slaan: I don't think overlay has an enable option, you'll probably have to split the video into 2 parts, the part you want to overlay and the part you don't, then concat them again
[20:43] <c_14> jthemovie: can you pastebin your complete command and the complete console output?
[20:43] <c_14> you might be able to turn of tls_verify if that's the problem
[20:44] <jthemovie> yeah this is what im looking for, as i need charles proxy in between to sniff this Secure HLS
[20:44] <jthemovie> it was working perfect with 1.10 version
[20:44] <jthemovie> i will give up pastebin
[20:45] <slaan> Mmm, okay c_14. But i have many many video to convert. I want to watermark this video each 10min of duration. I think this method could be hard :)
[20:45] <jthemovie> c_14
[20:46] <jthemovie> you got it here>> http://pastebin.com/Ziwa88MJ
[20:46] <c_14> slaan: limit the overlay video to 10min and then use eof_action
[20:46] <jthemovie> of course, the key has been changed and it won´t never play withtout cookie and some others http headers
[20:47] <jthemovie> but i wish you could find me this options in command line that will do tls_verify=false or something like that
[20:47] <jthemovie> because recompiling ffmeg from scratch is a long process on my single core cpu
[20:47] <slaan> Okay ! i check that, thx c_14 !
[20:48] <c_14> jthemovie: have you tried -tls_verify 0 ?
[20:49] <jthemovie> yeap, but doesn't change at all :/
[20:50] <jthemovie> i just did like this
[20:50] <jthemovie> ffmpeg -tls_verify 0 -re -i "https://beinsportnet4-lh.akamaihd.net/i
[20:51] <jthemovie> 4     {"tls_verify", "Verify the peer certificate",         http://ffmpeg.org/doxygen/trunk/tls_8c.html#ad12dce0a7bf9d908b172a28155b3d261(verify),    http://ffmpeg.org/doxygen/trunk/group__avoptions.html#ggabd75aa30eb8ad6387672df9a1fa79444ac068b8f6afeeb293a13aeabb5d702411, { .i64 = 0 }, 0, 1, .flags = http://ffmpeg.org/doxygen/trunk/tls_8c.html#af316c33cc298530f245e8b55330e86b5|http://ffmpeg.org/doxygen/trunk/tls_8c.html#a07484107e
[20:51] <jthemovie> i see it in the code , its part of AVOptions
[20:51] <jthemovie> line 84 in tlc.c
[20:53] <c_14> Why does your pastebin have the output from 2 ffmpeg versions?
[20:54] <c_14> aaah
[20:54] <c_14> Didn't see the pipe
[20:54] <jthemovie> :)
[20:56] <jthemovie> Even the static build are compiled with this TLS options :/
[20:57] <c_14> I'm not even sure what the 'fatal alert' is trying to tell me...
[20:58] <c_14> Also, it appears tls_verify is off by default
[20:58] <jthemovie> well it simple, ffmpeg run on a vps somewhere
[20:58] <jthemovie> its debian based
[20:58] <c_14> how about running curl in front
[20:58] <c_14> then piping that into ffmpeg
[20:59] <jthemovie> it send the request via charls proxy sniffing ssl
[20:59] <c_14> instead of `ffmpeg -re -i "url"' use curl -k "url" | ffmpeg -re -i pipe:0
[20:59] <jthemovie> i do have the ca.crt, may be i can use it somewhere in the command line?
[20:59] <c_14> you can pass -ca_file filename
[21:00] <jthemovie> ah yeah true, curl allow bypassing certificate
[21:00] <jthemovie> let metry
[21:09] <jthemovie> take time, i need to regerate token, propagate it to vps, modify cookie in charles :)
[21:21] <jthemovie> with the pipe:0 it still do the same
[21:21] <jthemovie> curl -lkL "https&." --ciphers DHE-RSA-AES256-SHA | ffmpeg -re -i pipe:0 -c:...
[21:22] <jthemovie> curl get the playlist with the fragment correctly though
[21:22] <c_14> What about just the `curl -kL "url" > /dev/null' ?
[21:22] <c_14> Does that also throw the error?
[21:22] <jthemovie> nope
[21:22] <jthemovie> curl request get it fine
[21:23] <jthemovie> with the ignore certificate (-k) options
[21:23] <jthemovie> wget too
[21:23] <jthemovie> wanna see the ouput?
[21:23] <jthemovie> via pastebin?
[21:24] <jthemovie> ii see on my proxy that it load the fist playlist
[21:24] <jthemovie> and try 2 or 3 fragments, then fail
[21:24] <c_14> aah, it's an m3u8 playlist
[21:25] <c_14> ye, curl won't help there since it'll only deal with the first link and not the subsequent ones
[21:25] <jthemovie> yeah its an HLS stream
[21:25] <jthemovie> alright, so may be i can try with -ca_file <filename>
[21:26] <jthemovie> charles public cert is public to everyone
[21:27] <c_14> try it
[21:30] <jthemovie> nope :(
[21:30] <jthemovie> it does recognize the parametr though as it don't stdout anything wrong abou it
[21:30] <jthemovie> but still the same
[21:30] <jthemovie> i have this swet feeling that tls_verify or ca_file are just simply ignoreed
[21:31] <jthemovie> and i'm not that brave to run gdb on it tonight :/
[21:33] <c_14> I have an inkling that the parameters are being used for the first url, but getting lost when it comes to parsing the hls and grabbing the subsequent urls
[21:34] <jthemovie> for checking my proxy requests
[21:34] <jthemovie> the request of the playlist just fail
[21:34] <jthemovie> not by 403 or 401, doesn´t reach the server i see
[21:34] <jthemovie> its a red cross on charles proxy
[21:35] <c_14> Is that with/without the curl and for the m3u file or for the first link therein?
[21:35] <c_14> And you said you (at some point) had an ffmpeg version it worked with?
[21:36] <jthemovie> yeah , just 1h ago
[21:36] <jthemovie> before i updagrade it
[21:36] <c_14> What version was that?
[21:36] <jthemovie> it was version stable 1.10-5 i guess
[21:36] <jthemovie> old version
[21:36] <c_14> Ok, that's rather old.
[21:36] <jthemovie> definitely
[21:36] <c_14> Was wondering if it might be a regression.
[21:37] <jthemovie> could you advise me some compilation flag to make sure i will compile it withtout ?
[21:37] <jthemovie> i just pull from git
[21:37] <jthemovie> is it disable-ssl ?
[21:38] <jthemovie> i see that this version is compiled with --enable-gnutls
[21:38] <c_14> If you disable ssl you won't be able to grab the tls links at all, I'd try linking against openssl instead of gnutls and hope it works with that.
[21:39] <jthemovie> do u mean that i simply remove the --enable-gnutls ?
[21:39] <jthemovie> not sure if i got u correctly sorry
[21:39] <c_14> Remove --enable-gnutls add --enable-openssl
[21:39] <jthemovie> ah ok get it
[21:40] <jthemovie> r so many server running tls yet?
[21:40] <jthemovie> mine is till in ssl v3
[21:43] <jthemovie> anway, thanks for you precious help c_14
[21:43] <c_14> http://disablessl3.com/
[21:44] <jthemovie> ur right, far from safe anymore
[21:44] <jthemovie> well then im gonna let it compile during the night i guess
[21:45] <jthemovie> really appreciate ur help again :) next time i will think twice before making such huge upgrade of version
[22:35] <sagax> hi all!
[22:35] <sagax> please, promt me
[22:36] <sagax> howto draw text in window when ffplay playing music track?
[22:36] <sagax> this work when ffplay playing video  https://ffmpeg.org/ffplay-all.html#drawtext-1
[22:36] <sagax> but, when i wish drawtext when ffplay fplaying audio?
[22:38] <kepstin-laptop> you'd have to add a video track, for example by using -filter_complex with the 'color' filter to create a blank black video.
[22:39] <sagax> hm, but i want to save a sound indicator
[22:40] <sagax> or something like that sound indicators
[22:40] <sagax> such as audio level indicator
[22:53] <c_14> sagax: something like showwaves or showspectrum?
[22:54] <sagax> yes
[22:56] <c_14> then use one of those filters and put the drawtext after in in the filterchain
[23:43] <sagax> thanks!
[23:43] <sagax> work, but only local file
[23:43] <c_14> hmm?
[23:43] <sagax> showing my command
[23:44] <sagax> ffplay -f lavfi 'amovie=/home/sagax/C7K:0/Eruku-Tatakai.mp3,asplit[out0],showwaves[out1]' -vf drawtext="fontfile=/home/sagax/.fonts/VeraMono.ttf:fontcolor="#ffffff":text='hi all'"
[23:44] <sagax> this work, but i try play sound stream
[23:44] <sagax> with link to online radio
[23:45] <sagax> and strim don't work
[23:46] <c_14> ffmpeg -i http://wheeeeeeee -filter_complex 'asplit[a1][a2];[a1]showwaves,drawtext=foobar[v]' -map '[v]' -f sdl /dev/null -map '[a2]' -f alsa default
[23:47] <sagax> hm, user ffmpeg
[23:48] <c_14> You can also pipe the output into your player of choice
[00:00] --- Tue Mar  3 2015


More information about the Ffmpeg-devel-irc mailing list